18.440 Midterm 2 Solutions, Fall 2009
1.
(a) P {X < a} = a2 for a ∈ (0, 1) and thus FX (a) = a2 . Differentiating
gives
2a a ∈ [0, 1]
fX (a) =
0 otherwise
(b) E[X] =
1
0 fX (x)xdx
=
1
2
0 2x dx
= 2/3.
(c) The pair (X, Y ) is uniformly distributed on the triangle
{(x, y) : x ∈ [0, 1], y ∈ [0, 1], x > y}. Thus, conditioned on X, Y is
uniform on [0, X], so E[Y |X] = X/2.
(d) Cov(X, Y ) = E[XY ] − E[X]E[Y ] = E[A1 A2 ] − E[X](E[1 − X]) =
1/4 − 2/9 = 1/36.
2.
(a) As derived in lecture and on problem sets, the collection of two
Poisson processes (the earthquake and the flood processes) may be
constructed equivalently by first taking a Poisson point process of
rate 2 + 3 = 5 and then independently declaring each point in the
process to be an earthquake with probability 3/5 (and a flood
otherwise). Thus, the number N of earthquakes before the first flood
satisfies P {N = k} = (3/5)k (2/5).
(b) One may recall that the sum of independent rate one exponentials is
a gamma distribution with α = 2 and λ = 1: the density is thus
f (t) =
e−λttα−1 λα /Γ[α] = e−t t t ≥ 0
.
0
otherwise
This can also be derived directly by letting X and Y denote the two
exponential random variables and writing
� t
� t
fX+Y (t) =
fX (x)fY (t − x)dx =
e−x e−(t−x) dx = te−t .
x=0
x=0
1
(c)
Cov(min{E, F, M }, M ) = Cov(min{E, F, M }, min{E, F, M })
+Cov(min{E, F, M }, M − min{E, F, M })
The memoryless property of exponentials implies that min{E, F, M }
and M − min{E, F, M } are independent, so this becomes
Var(min{E, F, M }). This is the variance of an exponential of rate 6,
which is 1/36.
3.
(a) E[X 2 Y 2 ] − E[XY ]2 = E[X 2 ]E[Y 2 ] − E[X]2 E[Y ]2 = E[X 2 ]E[Y 2 ] = 13 .
2 /2 et −1
(b) E[etX etY ] = E[etX ]E[etY ] = et
t
.
4.
100
60
(a) Cov(X, Y ) = Cov( 60
, X ) = 20.
i=1 Xi ,
j=41 Xj ) =
i=41 Cov(X
√i i
Since Var(X) = Var(Y ) = 60, we have ρ(X, Y ) = 20/ 60 · 60 = 1/3.
(b) Let Z = 100
i=61 Xi . Then X and Z are independent with joint
2
2
density f (a, b) = √2π1√60 e−a /120 √2π1√40 e−b /80 . Conditioning on
X + Z = x restricts us to the line a + b = x, so the conditional
density for X will be (for some constant C)
2
2 /80
f (a) = Cf (a, x − a) = Ce−a /120 e−(x−a)
R∞
where C is chosen so that −∞ f (a)da = 1.
,
5.
the ith year, then
(a) If Xi is the multiplicative factor during
u
E[Xi ] = 2 · .4 + .5 · .6 = 1.1 and E Xi = 1.1100 ∼ 13780.6.
(b) We need to get at least 50 “up” steps. Mean number is 40,√variance is
100(.6 · .4) = 24. So, very roughly, the chance is 1 − Φ(10/ 24) ∼ .96.
Despite high expectation, investment usually loses money.
2
MIT OpenCourseWare
http://ocw.mit.edu
18.440 Probability and Random Variables
Spring 2014
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.